Objective To investigate the attributable risk (AR) of Acinetobacter baumannii (AB) infection in criti- cally ill patients. Methods A multicenter retrospective cohort study was conducted among adult patients in intensive care unit (ICU). Patients with AB isolated from sterile body fluid and confirmed with AB infection in each center were selected as the infected group. According to the matching criteria that patients should be from the same period, in the same ICU, as well as with similar APACHE Ⅱ score (±5 points) and primary diagnosis, patients who did not infect with AB were selected as the non-infected group in a 1:2 ratio. The AR was calculated. Results The in-hospital mortality of patients with AB infection in sterile body fluid was 33.3%, and that of non-infected group was 23.1%, with no statistically significant difference between the two groups (P=0.069). The AR was 10.2% (95%CI: -2.3%-22.8%). There is no statistically significant difference in mortality between non-infected patients and infected patients from whose blood, cerebrospinal fluid and other specimen sources AB were isolated (P>0.05). After infected with AB, critically ill patients with the major diagnosis of pulmonary infection had the highest AR. There was no statistically significant difference in mortality between patients in the infected and non-infected groups (P>0.05), or between other diagnostic classifications. Conclusion The prognosis of AB infection in critically ill patients is highly overestimated, but active healthcare-associated infection control for AB in the ICU should still be carried out.